<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=ISO-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7651.28">
<TITLE>RE: [Haifux] talk offer: tapping into the fountaion of CPUs</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>Hi,<BR>
<BR>
Certainly !<BR>
<BR>
Yossi Arbel<BR>
<BR>
-----Original Message-----<BR>
From: haifux-bounces@haifux.org on behalf of Muli Ben-Yehuda<BR>
Sent: Wed 1/2/2008 12:20 PM<BR>
To: haifux@haifux.org<BR>
Subject: [Haifux] talk offer: tapping into the fountaion of CPUs<BR>
<BR>
Would there be interest from the esteemed list members in hearing the<BR>
following paper, to be presented at ASPLOS '08?<BR>
<BR>
Tapping into the Fountain of CPUs---On Operating System Support for<BR>
Programmable Devices, by Yaron Weinsberg, Danny Dolev, Tal Anker, Muli<BR>
Ben-Yehuda, Pete Wyckoff.<BR>
<BR>
Abstract: The constant race for faster and more powerful CPUs is<BR>
drawing to a close. No longer is it feasible to significantly increase<BR>
the speed of the CPU without paying a crushing penalty in power<BR>
consumption and production costs. Instead of increasing single thread<BR>
performance, the industry is turning to multiple CPU threads or cores<BR>
(such as SMT and CMP) and heterogeneous CPU architectures (such as the<BR>
Cell Broadband Engine). While this is a step in the right direction,<BR>
in every modern PC there is a wealth of untapped compute<BR>
resources. The NIC has a CPU; the disk controller is programmable;<BR>
some high-end graphics adapters are already more powerful than host<BR>
CPUs. Some of these CPUs can perform some functions more efficiently<BR>
than the host CPUs. Our operating systems and programming<BR>
abstractions should be expanded to let applications tap into these<BR>
computational resources and make the best use of them.<BR>
<BR>
Therefore, we propose the Hydra framework, which lets application<BR>
developers use the combined power of every compute resource in a<BR>
coherent way. Hydra is a programming model and a runtime support layer<BR>
which enables utilization of host processors as well as various<BR>
programmable peripheral devices' processors. We present the framework<BR>
and its application for a demonstrative use-case, as well as provide a<BR>
thorough evaluation of its capabilities. Using Hydra we were able to<BR>
cut down the development cost of a system that uses multiple<BR>
heterogenous compute resources significantly.<BR>
<BR>
Cheers,<BR>
Muli<BR>
_______________________________________________<BR>
Haifux mailing list<BR>
Haifux@haifux.org<BR>
<A HREF="http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux">http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux</A><BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>